[plug] Linux/OSS based network knowledge packages?

[Denis Brown]

Dear PLUG list members,

Network diagramming / database software, preferrably open source, to run
on Linux that someone has actually used...   The Windows-based netViz
(www.netviz.com) dr.diagram book suite looks ideal.   Google has several
suggestions but Scotty/Tkined, MapNet, Geotraceman do not do what I'm
after at a cabling level, that I can see.  They are too high-level.

My pipe dream is a system for capturing and reguritating physical network
data, for easy troubleshooting.   For example a fibre cable between two
buildings has certain physical characteristics - it has been puchased and
installed by identifiable entities (for asset tracking purposes), it
consists of N circuits, each circuit carries one or more subnets, each end
terminates at some site, building, level, room number, rack number,
patchpanel ID, connector IDs and possibly also listing connector type.
This cable may appear on one or more site drawings, so the relevant
drawing number(s) also need to be recorded.

Each circuit in the cable, at each end, is patched to some other cable /
equipment - such as a switch or router, or possibly a straight-through
fibre patch to another between-building or between-floor cable.   And so
on.   Each item of equipment is then recorded... a switch has N ports, a
model & serial number, an admin username/password and IP adddress (if
managable switch)[1], what type of ports (fibre, UTP, ...), what subnet(s)
does it handle, etc.   In a database model there would be tables for cable
runs, various types of equipment, etc.

On the "output" side of this collection of data it should then be possible
to query by subnet -- find all cables, switches, maybe even patchleads,
media convertors, etc that pertain to it.   Query by building/level/room
etc to find out what would be affected if power was lost to this location,
and so on.   Query by equipment ID - if I want to manage this switch, what
is its username/password or where do I find it on the 'net.

Netviz looks ideal and maybe I should just look into costs and whether it
will run on Crossover Office, for example.   But I thought I'd ask here
first for any other solutions that people find handy in the real world,
especially if they're open source.   Scale-wise we're talking several
sites, several buildings, most multi-floor and a userbase (number of
network appliances, printers, PCs, servers) in the high hundreds.

TIA,
Denis

[1] obviously we'd be talking a secure database here!!